Claims resolved in minutes — Vision AI analyses, decides, executes

The system monitors all claims channels: email inbox with photo/video attachments, customer portal forms, website chat with file upload. AI classifies the claim type (transport claim, product damage, order discrepancy, short delivery, wrong item) and retrieves context: identifies the customer, loads the related order from ERP, checks claims history (whether the customer has raised similar claims before), retrieves carrier data and delivery status.

Vision AI analyses the image with over 98% accuracy: identifies the damage type (mechanical, flooding, deformation, crack, missing component), locates it on the product or packaging, assesses the scale, distinguishes product damage from packaging damage (critical for carrier liability). The agent verifies that the product in the photo matches what was ordered, cross-checks with transport data (whether the parcel had damage markings at delivery), and detects fraud attempts (same photo from a previous claim, edited image, inconsistency with packaging). Output: validity status + action proposal + confidence score.

For valid claims with confidence above the company policy threshold: AI autonomously creates the claims document in ERP (return document, credit note, replacement order) and sends confirmation to the customer in their language. High-value, ambiguous or red-flagged claims go to the service team with a ready AI recommendation and full visual documentation — the person does not start from scratch, they verify the proposal. Full audit log of every decision with preserved images for compliance and potential carrier disputes.

AI resolves autonomously claims that satisfy three conditions simultaneously: clear damage type with confidence above 95%, amount below the autonomous decision threshold (configurable per product category) and no red flags in the customer history. Everything outside this goes to the service team with a ready recommendation — visual analysis, action proposal, customer context and suggested response. Every autonomous AI decision is in the audit log with preserved image, model ID and confidence score. The customer can raise an appeal at any time — the case then goes to a human with full context. AI proposes. Human decides. ERP remains your system of truth.